The outcome in acute myeloid leukemia (AML) varies widely between individual patients. Given this heterogeneity, there is long-standing interest in identifying predictors of therapeutic response. However, even complex models that integrate many of these established factors currently have a predictive accuracy that is closer to a coin flip than certainty, documenting the need for the delineation of novel, critical biological feature to refine disease-risk classification and allow better-tailored, risk-stratified treatment approaches. Our recent data suggest that the very late antigen-4 (VLA-4) integrin may define such a marker. Specifically, among 216 participants of the Children's Oncology Group (COG)-AAML03P1 trial, we found that high VLA-4 expression independently predicted a lower risk of relapse (RR) and superior disease-free survival (DFS); this effect was particularly pronounced in standard-risk patients, i.e. those diverse patients that often lack identifiable cytogenetic/molecular risk factors. So far, the biological basis for the observed association between VLA-4 expression and outcome is unknown. We now propose to validate our previous findings on the role of VLA-4 as prognostic marker in the COG-AAML0531 study, a phase 3 trial on >1,000 pediatric patients with newly diagnosed AML, and to define its quantitative contribution to the accuracy of prediction models relative to other, established markers. We will also test our hypothesis that high VLA-4 expression is associated with a specific expression profile of adhesion molecules and extracellular matrix (ECM) proteins, as suggested by our preliminary studies, as well as distinct responses to VLA-4 ligands that, ultimately, result in reduced chemoresistance and improved outcome and provide a biological basis for the observed predictive ability of VLA-4.] Specifically, we will: 1) Determine whether high VLA-4 expression on diagnostic AML blast cells predicts a reduced RR and superior DFS in pediatric patients with newly diagnosed AML enrolled in COG-AAML0531; 2) Define the expression profiles of adhesion molecules and ECM proteins in the COG-AAML0531 trial cohort and determine their association with VLA-4 expression and predictive role for outcome after induction and consolidation chemotherapy; and 3) Compare and validate the gene expression signatures in a targeted set of biospecimens with high and low VLA-4 expression at baseline and in response to physiological VLA-4 ligands. These studies will further define and validate the role of VLA-4 and other adhesion molecules and ECM proteins as predictors of therapeutic response in pediatric AML. Our proposed research may identify adhesion molecule expression patterns that contribute to the inter-patient variability in response to chemotherapy and provide the rationale for personalized, risk-adapted therapeutic approaches, a major advancement over current treatment strategies. Furthermore, beyond the delineation of the prognostic role of such proteins, our studies may discover novel targets for therapeutic drug development and provide the rationale for mechanistic studies aimed at understanding their functional role in leukemogenesis.